Explosion of Open-Source Models Outpaced by Industry's Closed-Source Dominance While Training Costs, U.S. Investment, and AI Adoption Skyrocket
-
There has been an explosion in the number of open-sourced foundation models released, though closed-source models still outperform them.
-
Industry, especially Google and OpenAI, dominates in developing and releasing foundation models, far outpacing academia.
-
The computational cost to train models has skyrocketed, with models like Gemini Ultra costing over $190 million.
-
The U.S. leads in originating notable ML models and in AI private investment, with over 8 times more investment than China.
-
More companies are adopting AI in their business, while individuals worry about AI's impact on jobs, especially younger and wealthier respondents.